SEBI’s Triple Bet: Broker Rules, Tokenised Bonds and an AI Sandbox

Oct 27

5 min read

A Market That Has Outgrown Its Plumbing

India now counts more than fifteen crore investors and a market capitalisation of about ₹472 lakh crore, yet until January this year its stockbrokers were governed by a rulebook written in 1992, and it is against this gap between the size of the market and the age of its rules that the latest proposals of the Securities and Exchange Board of India (SEBI) should be read.

Speaking in Mumbai on 24 October 2026, SEBI Chairman Tuhin Kanta Pandey outlined graded compliance for brokers according to their scale, client exposure and reliance on technology, secondary-market trading of tokenised corporate bonds under the next phase of Demat 2.0, and a framework for responsible artificial intelligence supported by a regulatory sandbox, which together mark a shift from rule-by-checklist towards rule-by-risk.

This commentary argues that the direction is correct and deserves support, but that success depends on three conditions not yet settled: that lighter compliance for small brokers does not mean weaker protection for small investors, that tokenised bonds find genuine buyers and sellers, and that the AI sandbox becomes a disciplined testing ground rather than a shelter from accountability. Each reform is examined in turn, followed by the main objections and by recommendations that match these three conditions.

Why the Direction Deserves Support

Proportionality for brokers is a matter of fairness, not leniency

A small broker serving a few hundred clients in a district town and a national platform serving millions do not pose the same risk, yet both have long faced much the same returns and inspections. The SEBI (Stock Brokers) Regulations, 2026, notified on 7 January 2026, began to correct this by consolidating scattered circulars and creating a category of Qualified Stock Brokers with heavier duties, and graded compliance would carry the same principle of proportion into daily supervision.

Uniform fixed costs also weigh most heavily on small firms, pushing them to merge or exit and concentrating market share and operational risk in a few large platforms, so that matching obligations to risk, together with SEBI’s plan for a single digital inspection workflow, points towards smarter supervision rather than lighter supervision.

Tokenised bonds address a real and long-standing weakness

India’s corporate bond market exceeds ₹60 trillion but equals only about 16 to 17 per cent of GDP, against roughly 79 per cent in South Korea, and SEBI Whole-Time Member Amarjeet Singh noted in August 2026 that of nearly 33,000 outstanding instruments only 400 to 500 trade on a typical day. With about 98 per cent of bonds privately placed and usually held to maturity, a retail investor who must sell early often finds no fair buyer, and a market that cannot discover prices cannot lower the cost of borrowing.

The first phase of Demat 2.0 has shown that the technology works within India’s legal structure, since REC Limited raised ₹500 crore on 7 September 2026 against bids of nearly ₹796 crore with same-day settlement and listing, while the depository remains the authoritative record of ownership, keeping tokenised bonds inside ordinary demat accounts and ordinary know-your-customer rules.

Guardrails for artificial intelligence are needed before harm, not after it

Artificial intelligence is already used by exchanges, brokers and fund houses for customer service, surveillance and order execution, and SEBI’s consultation paper of 20 June 2025 set out the principles of accountable management, explainable models, testing before deployment and disclosure to investors, while the proposed tiered framework, with human oversight and emergency “kill switches,” builds faithfully on that foundation and on the work of the International Organization of Securities Commissions.

The need becomes clear when the retail investor is considered, since SEBI’s studies found that about 91 per cent of individual derivatives traders lost money in 2024–25 and 87.7 per cent in 2025–26, while only 36 per cent of investors reported moderate to high market knowledge; in such a setting a poorly governed algorithm could harm inexperienced users at great speed, and a sandbox offers innovators clarity while giving the regulator a supervised view of how these tools behave.

Building the Plumbing to Match the Market

A market of fifteen crore investors cannot be served by rules designed for the early 1990s, and the earlier conclusion therefore stands: the package deserves support, provided that the three conditions set out at the start are met in the detailed rules still to be published.

First, graded compliance should be introduced through a public consultation that lists which obligations will vary and which will not, with client-asset protection, balance reporting and complaint timelines kept uniform for every broker, and with objective, published criteria for placing firms in tiers.

Second, secondary trading of tokenised bonds should begin alongside the market-making framework rather than ahead of it, and SEBI should publish pilot data on volumes, spreads and settlement failures, and settle in its rules whether code or trust deed prevails, before extending the model to lower-rated issuers or to retail investors at scale.

Third, the AI sandbox should have clear entry criteria, caps on users and exposure, a fixed testing period, compulsory disclosure to investors and a duty to shut systems down at once, while the final AI guidelines, pending since June 2025, should be notified without further delay.

If these steps are taken, the October 2026 announcements may be remembered as a coherent effort to modernise India’s market infrastructure; if they are not, the same proposals may yield lighter broker oversight, a faster but still empty bond market and an experimental space without accountability, and which outcome prevails will be decided by the precision of the rules that follow.

Sources

Securities and Exchange Board of India. “Consultation Paper on Guidelines for Responsible Usage of AI/ML in Indian Securities Markets.” June 20, 2025. https://www.sebi.gov.in/reports-and-statistics/reports/jun-2025/consultation-paper-on-guidelines-for-responsible-usage-of-ai-ml-in-indian-securities-markets_94687.html.
Singh, Amarjeet. “Speaking Notes of Shri Amarjeet Singh, WTM, SEBI at ASSOCHAM’s 9th National Summit & Awards on Corporate Bond Market.” Securities and Exchange Board of India, August 20, 2026. https://www.sebi.gov.in/sebi_data/attachdocs/aug-2026/1787219973644.pdf.
REC Limited. “Press Release: REC Limited Successfully Issued India’s First Pilot Issue on Tokenized Corporate Bonds under SEBI Regulatory Sandbox Framework.” Filing to the National Stock Exchange of India, September 7, 2026. https://nsearchives.nseindia.com/corporate/RECLTD_07092026192505_Signed_Press_Release_07092026.pdf.
International Organization of Securities Commissions. The Use of Artificial Intelligence and Machine Learning by Market Intermediaries and Asset Managers: Final Report. Madrid: IOSCO, 2021. https://www.iosco.org/library/pubdocs/pdf/IOSCOPD658.pdf.
“SEBI Plans New Stockbroker Rules, Tokenised Bonds and AI Sandbox.” Times Now, October 2026. https://www.timesnownews.com/business-economy/markets/sebi-plans-new-stockbroker-rules-tokenised-bonds-and-ai-sandbox-article-156397262.
Rana, Shipra. “SEBI’s Gift for Stock Brokers: Compliance 2.0, AI Rules in Focus.” Free Press Journal, October 24, 2026. https://www.freepressjournal.in/business/sebis-gift-for-stock-brokers-compliance-20-ai-rules-in-focus.
Anupama VM. “SEBI Outlines Capital Market Reforms with Listing, Bond and AI Plans as Market Capitalisation Reaches ₹472 Lakh Crore.” 5paisa, October 15, 2026. https://www.5paisa.com/blog/sebi-capital-market-reforms-listing-bonds-ai.
“SEBI F&O Loss Study Explained: Why 9 in 10 Retail Traders Lost ₹91,685 Crore in FY26.” Open, 2026. https://openthemagazine.com/business/sebi-fo-loss-study-explained-why-9-in-10-retail-traders-lost-91685-crore-in-fy26.